You Only Look Once, But the Parts Keep Moving: YOLO-Based Workpiece Pose Classification for Aerodynamic Part Feeding

Research output: Chapter in book/report/conference proceedingConference contributionResearchpeer review

Authors

View graph of relations

Details

Original languageEnglish
Title of host publication2025 IEEE 21st International Conference on Automation Science and Engineering, CASE 2025
PublisherIEEE Computer Society
Pages153-158
Number of pages6
ISBN (electronic)9798331522469
ISBN (print)979-8-3315-2247-6
Publication statusPublished - 17 Aug 2025
Event21st IEEE International Conference on Automation Science and Engineering, CASE 2025 - Los Angeles, United States
Duration: 17 Aug 202521 Aug 2025

Publication series

NameIEEE International Conference on Automation Science and Engineering
ISSN (Print)2161-8070
ISSN (electronic)2161-8089

Abstract

Flexible part feeding is a key challenge in modern automated production, where increasing uncertainties, shorter product life cycles, and cost pressures require adaptable solutions. Aerodynamic part feeding systems, which use controlled air jets to manipulate workpieces, offer a retooling-free alternative to traditional vibratory bowl feeders. To ensure precise workpiece handling, reliable pose classification is essential. This paper presents a machine learning-based framework for classifying workpiece poses using a class of convolutional neural networks (CNNs) called YOLO and an industrial camera. Instead of relying on manually labeled real-world images-which would introduce machine downtimes and increased setup efforts-the proposed method trains CNNs exclusively on synthetic datasets. Artificial images of workpieces in various poses are generated from CAD models using the open-source rendering engine Blender. Multiple CNN architectures are trained and evaluated, achieving a classification precision exceeding 95 % for most workpieces when tested on real workpiece images. The results demonstrate that the approach enables accurate and efficient workpiece pose classification without the need for labor-intensive dataset creation. While developed for aerodynamic part feeding, the proposed method is applicable to a wide range of industrial scenarios requiring automated workpiece orientation classification.

ASJC Scopus subject areas

Cite this

You Only Look Once, But the Parts Keep Moving: YOLO-Based Workpiece Pose Classification for Aerodynamic Part Feeding. / Shieff, Dasha; Akchi, Mohamed; Raatz, Annika.
2025 IEEE 21st International Conference on Automation Science and Engineering, CASE 2025. IEEE Computer Society, 2025. p. 153-158 (IEEE International Conference on Automation Science and Engineering).

Research output: Chapter in book/report/conference proceedingConference contributionResearchpeer review

Shieff, D, Akchi, M & Raatz, A 2025, You Only Look Once, But the Parts Keep Moving: YOLO-Based Workpiece Pose Classification for Aerodynamic Part Feeding. in 2025 IEEE 21st International Conference on Automation Science and Engineering, CASE 2025. IEEE International Conference on Automation Science and Engineering, IEEE Computer Society, pp. 153-158, 21st IEEE International Conference on Automation Science and Engineering, CASE 2025, Los Angeles, California, United States, 17 Aug 2025. https://doi.org/10.1109/CASE58245.2025.11163971
Shieff, D., Akchi, M., & Raatz, A. (2025). You Only Look Once, But the Parts Keep Moving: YOLO-Based Workpiece Pose Classification for Aerodynamic Part Feeding. In 2025 IEEE 21st International Conference on Automation Science and Engineering, CASE 2025 (pp. 153-158). (IEEE International Conference on Automation Science and Engineering). IEEE Computer Society. https://doi.org/10.1109/CASE58245.2025.11163971
Shieff D, Akchi M, Raatz A. You Only Look Once, But the Parts Keep Moving: YOLO-Based Workpiece Pose Classification for Aerodynamic Part Feeding. In 2025 IEEE 21st International Conference on Automation Science and Engineering, CASE 2025. IEEE Computer Society. 2025. p. 153-158. (IEEE International Conference on Automation Science and Engineering). doi: 10.1109/CASE58245.2025.11163971
Shieff, Dasha ; Akchi, Mohamed ; Raatz, Annika. / You Only Look Once, But the Parts Keep Moving : YOLO-Based Workpiece Pose Classification for Aerodynamic Part Feeding. 2025 IEEE 21st International Conference on Automation Science and Engineering, CASE 2025. IEEE Computer Society, 2025. pp. 153-158 (IEEE International Conference on Automation Science and Engineering).
Download
@inproceedings{6dbcb3c03fbd470d95421b23110ad481,
title = "You Only Look Once, But the Parts Keep Moving: YOLO-Based Workpiece Pose Classification for Aerodynamic Part Feeding",
abstract = "Flexible part feeding is a key challenge in modern automated production, where increasing uncertainties, shorter product life cycles, and cost pressures require adaptable solutions. Aerodynamic part feeding systems, which use controlled air jets to manipulate workpieces, offer a retooling-free alternative to traditional vibratory bowl feeders. To ensure precise workpiece handling, reliable pose classification is essential. This paper presents a machine learning-based framework for classifying workpiece poses using a class of convolutional neural networks (CNNs) called YOLO and an industrial camera. Instead of relying on manually labeled real-world images-which would introduce machine downtimes and increased setup efforts-the proposed method trains CNNs exclusively on synthetic datasets. Artificial images of workpieces in various poses are generated from CAD models using the open-source rendering engine Blender. Multiple CNN architectures are trained and evaluated, achieving a classification precision exceeding 95 % for most workpieces when tested on real workpiece images. The results demonstrate that the approach enables accurate and efficient workpiece pose classification without the need for labor-intensive dataset creation. While developed for aerodynamic part feeding, the proposed method is applicable to a wide range of industrial scenarios requiring automated workpiece orientation classification.",
author = "Dasha Shieff and Mohamed Akchi and Annika Raatz",
note = "Publisher Copyright: {\textcopyright} 2025 IEEE.; 21st IEEE International Conference on Automation Science and Engineering, CASE 2025, CASE 2025 ; Conference date: 17-08-2025 Through 21-08-2025",
year = "2025",
month = aug,
day = "17",
doi = "10.1109/CASE58245.2025.11163971",
language = "English",
isbn = "979-8-3315-2247-6",
series = "IEEE International Conference on Automation Science and Engineering",
publisher = "IEEE Computer Society",
pages = "153--158",
booktitle = "2025 IEEE 21st International Conference on Automation Science and Engineering, CASE 2025",
address = "United States",

}

Download

TY - GEN

T1 - You Only Look Once, But the Parts Keep Moving

T2 - 21st IEEE International Conference on Automation Science and Engineering, CASE 2025

AU - Shieff, Dasha

AU - Akchi, Mohamed

AU - Raatz, Annika

N1 - Publisher Copyright: © 2025 IEEE.

PY - 2025/8/17

Y1 - 2025/8/17

N2 - Flexible part feeding is a key challenge in modern automated production, where increasing uncertainties, shorter product life cycles, and cost pressures require adaptable solutions. Aerodynamic part feeding systems, which use controlled air jets to manipulate workpieces, offer a retooling-free alternative to traditional vibratory bowl feeders. To ensure precise workpiece handling, reliable pose classification is essential. This paper presents a machine learning-based framework for classifying workpiece poses using a class of convolutional neural networks (CNNs) called YOLO and an industrial camera. Instead of relying on manually labeled real-world images-which would introduce machine downtimes and increased setup efforts-the proposed method trains CNNs exclusively on synthetic datasets. Artificial images of workpieces in various poses are generated from CAD models using the open-source rendering engine Blender. Multiple CNN architectures are trained and evaluated, achieving a classification precision exceeding 95 % for most workpieces when tested on real workpiece images. The results demonstrate that the approach enables accurate and efficient workpiece pose classification without the need for labor-intensive dataset creation. While developed for aerodynamic part feeding, the proposed method is applicable to a wide range of industrial scenarios requiring automated workpiece orientation classification.

AB - Flexible part feeding is a key challenge in modern automated production, where increasing uncertainties, shorter product life cycles, and cost pressures require adaptable solutions. Aerodynamic part feeding systems, which use controlled air jets to manipulate workpieces, offer a retooling-free alternative to traditional vibratory bowl feeders. To ensure precise workpiece handling, reliable pose classification is essential. This paper presents a machine learning-based framework for classifying workpiece poses using a class of convolutional neural networks (CNNs) called YOLO and an industrial camera. Instead of relying on manually labeled real-world images-which would introduce machine downtimes and increased setup efforts-the proposed method trains CNNs exclusively on synthetic datasets. Artificial images of workpieces in various poses are generated from CAD models using the open-source rendering engine Blender. Multiple CNN architectures are trained and evaluated, achieving a classification precision exceeding 95 % for most workpieces when tested on real workpiece images. The results demonstrate that the approach enables accurate and efficient workpiece pose classification without the need for labor-intensive dataset creation. While developed for aerodynamic part feeding, the proposed method is applicable to a wide range of industrial scenarios requiring automated workpiece orientation classification.

UR - http://www.scopus.com/inward/record.url?scp=105018334427&partnerID=8YFLogxK

U2 - 10.1109/CASE58245.2025.11163971

DO - 10.1109/CASE58245.2025.11163971

M3 - Conference contribution

AN - SCOPUS:105018334427

SN - 979-8-3315-2247-6

T3 - IEEE International Conference on Automation Science and Engineering

SP - 153

EP - 158

BT - 2025 IEEE 21st International Conference on Automation Science and Engineering, CASE 2025

PB - IEEE Computer Society

Y2 - 17 August 2025 through 21 August 2025

ER -

By the same author(s)